08 — Platform & Product Roadmap (18 months)
The platform exists to make cohort delivery + evidence reporting cheaper, faster, and more rigorous. Every feature must pass the test: does this reduce cost-to-deliver, or increase confidence factor, or both? If neither, it is descoped.
Horizons
- Now (0–3 months) — finish what is in flight; harden Sprint 5 dashboard
- Next (3–9 months) — fewer, larger bets that unlock S3
- Later (9–18 months) — bets that unlock multi-region (S4)
Now (0–3 months)
| Item | Build / Buy / Wait | Notes |
|---|---|---|
| Hybrid dashboard hardening (Sprint 5 §04) — uptime, latency, mobile | Build | Already in flight; finish |
| Evidence Room v1 — claim → query hash → snapshot (Sprint 8 §08) | Build | Moat-critical |
| Coach roster + utilisation report (feeds KPI #7) | Build | Internal tool, simple |
| Safeguarding case log | Buy | Existing safeguarding case-management SaaS; DSL-approved |
| Weekly scorecard auto-generation | Build (small) | Replaces manual deck Friday mornings |
Next (3–9 months)
| Item | Build / Buy / Wait | Notes |
|---|---|---|
| Cohort telemetry → SROI rollup automation (Sprint 5 §07) | Build | Removes spreadsheet-stitching from Evidence team |
| Coach matching v2 (rubric + availability + utilisation) | Build | Internal; uplift to Sprint 3 §02 |
| Multi-tenant client dashboard with role-based access | Build | Required for S3 enterprise accounts |
| Translation/localisation workflow for materials | Buy | Off-the-shelf LMS-adjacent tool + human review |
| Mobile participant experience (workbook + nudges) | Build (PWA) | Already on a PWA stack; minimal cost |
| Coach CPD tracker | Buy | Off-the-shelf if cheap; build only if integration cost > buy |
| Data residency split (UK / EU) | Build | Required for §05 wave 4 |
Later (9–18 months)
| Item | Build / Buy / Wait | Notes |
|---|---|---|
| Predictive cohort RAG (early warning from telemetry) | Build (carefully) | Only if we can ship with calibration evidence — no black-box scoring |
| External Evidence API for clients' BI tools | Build | Differentiator at S4; gated by Evidence Room v2 |
| AI-assisted facilitator prep (using Lovable AI gateway) | Build | Always assistive, never autonomous in delivery |
| Regional client portal localisation | Build | Follows §05 wave 3+ |
| Partner portal (Tier B) — pipeline, materials, reporting | Build (light) | Only when ≥ 3 active Tier B partners |
| Open public methodology browser (read-only Evidence Room slice) | Wait | Compelling moat play; gate on assurance maturity |
Build / buy / wait rubric
Build when:
- It touches our moat (evidence, coach quality, safeguarding)
- No mature SaaS exists at acceptable data/safety posture
- Internal workflow + small surface area (we can ship in < 6 weeks)
Buy when:
- Commodity workflow (CRM, finance, HRIS, case management)
- Vendor has strong security posture and data residency options
- Total cost of ownership over 3 years < build cost
Wait when:
- We do not yet know the right shape (no two clients agree)
- The cost of being wrong (rework, data migration) is high
- It is a "nice-to-have" without a binding constraint behind it
Anti-patterns
- Shipping AI features into delivery flow without a calibration story (Sprint 8 §05)
- Building a generic "platform" before we know which 3 workflows actually compound
- Buying a tool because it demos well, before mapping it to a workflow we already do manually
- Letting Engineering own roadmap prioritisation (it belongs to COO + Heads of Delivery, Coaching, Evidence)
Roadmap governance
- Roadmap reviewed monthly at MBR (Sprint 9 §04); reprioritised quarterly at QSR
- Every shipped feature has a one-line success metric agreed BEFORE build starts
- Anything not used by 60% of intended users within 90 days goes on a kill / fix list
- No more than 3 "Now" items concurrent; no more than 5 "Next" items committed